456 METALLURGY OF CAST IRON.
had a projection cast on the upper end exactly two feet from the bottom of the mould, which was formed by the bottom plate to measure contraction by. The three bars were poured by one runner in the center of the three moulds, the iron dropping from the top. I made these three bars two feet long, so as to give a greater length than was in the one' foot long by one-half inch square bars, to better detect any difference that might exist in the contraction of the bars due to variation in the 'l temper ' * of the sand. When these bars were measured, no difference could be found in their contraction — a further proof of the necessity of using a bar larger than one-half inch square or round to show the true contraction of an iron. I also made tests with one and one-eighth inch round bars cast flat, but did not find that the radical variation which existed in the '' temper '' of the sand made any difference in the length of their contraction. Previous to these tests, I also made some in our foundry in the presence of E. Duqne Estrada, M. E., of Pittsburg, a member of the American Society of Mechanical Engineers' Testing Committee, to learn whether degrees in fluidity of iron would affect the contraction of large-sized test bars or thick castings. To test this point, two bars two inches square and forty-eight inches long were moulded together in the same mould. One was poured with the metal as '' hot'' as could be obtained from the cupola, and the other with the same ladle cooled down to pour the metal as '' dull'' as possible and still obtain a full-run bar. Two sets of these experiments were made, but no difference was found in their contraction. The fact of there being no visible difference in the contraction of the two-inchameter and made a dry sand mould, using" a piece \